Developing a Subregion Model Tool for the Southern

Developing a Subregion Model Tool
for the Southern California
Association of Governments
Jim Lam, Caliper Corporation
Guoxiong Huang, SCAG
Steve Smith, SANBAG
TRB Tools of the Trade Conference, Sept. 24, 2010
Outline
•
•
•
•
•
Goal of the Subregion Tool
Subregion Tool Process
Subregion Models
Subregion Tool Demo
Model Application – San Bernardino
County
• Future Work
Goal of the Subregion Tool
• Current SCAG Regional Model is 4000+
zones, 100,000+ links, and takes 24
hours to run
• SCAG region contains 170+ cities, 6
counties, and 17 subregions, many with
modeling needs
• Very expensive and time consuming to
build custom subregion models for each
• Tool designed to quickly use the
regional model to create subregion
models
• Models designed to focus on subregion,
but be compatible with Regional Model
Typical Subregion Model Development
• Subregion defined
• TAZs and network inside subregion
disaggregated
• TAZs and network outside subregion
aggregated
• Network, TAZs, demographics, other input
manually re-configured
• Model inside subregion much more accurate
• Model outside subregion much less accurate
• Model is calibrated and validated for area
inside subregion and model is ignored
outside subregion
Subregion Tool Process
• Main goal: automate and standardize
conversion
• Subregion Conversion Steps
– Define master subregion GIS layer with
disaggregation inside subarea
– Run a conversion utility which
• Auto-disaggregates and aggregates TAZs,
demographics and other table inputs
• Auto-disaggregates and aggregates matrix inputs
• Auto converts network and creates new centroid
connectors
– Run Subregion version of Regional Model
• Similar to Regional Model with exception in
aggregation areas to ensure consistency
Inside Subregion
• TAZ attributes re-calculated based upon
population/employment ratios, or input
from subregion agency
• Centroid connectors redefined
automatically, or based on previous
definitions
• Additional network links automatically
merged in
Inside Subregion Example
Before Disaggregation
After Disaggregation
Outside Subregion
• Centroid connectors preserved, but
aggregated to higher lever
• Zonal aggregation models added to
– Account for aggregation bias
– Account for trip loss during aggregation
Centroid Aggregation Outside Subregion
What is Aggregation Bias?
• Model estimated based upon defined zone
system
• Model evaluated based on different set of
zones
• Solution: Keep original zone locations
• For each aggregate zone, pick a
representative zone
Zonal Aggregation Models
• Intra-Superzone Trips Model
• To account for underassignment of trips
This Trip
Becomes an Intrazonal Trip
Zonal Aggregation Models
• Intra-Superzone Trips Model
• Take intra-superzone trips and load
them onto the networks using original
connectors
• Combine with aggregate assigned trips
Intra-Superzone Assignment Example
Aggregate OD Matrix
Correspondence Table
Intra-Superzone Trip
Inter-Superzone Trip
Disaggregate Intra-Superzone Trip
Demo of Subregional Model
Verification Results-San Bernardino
Assignment Speed, VMT, VHT, Delay
Time Period
Regional Model
Average Speed (mph), ALL
Vehicle Miles Traveled ('000), ALL
Vehicle Hours Traveled ('000), ALL
Vehicle Hours Delay ('000), ALL
Subregion Model
Average Speed (mph), ALL
Vehicle Miles Traveled ('000), ALL
Vehicle Hours Traveled ('000), ALL
Vehicle Hours Delay ('000), ALL
Difference
Average Speed (mph), ALL
Vehicle Miles Traveled ('000), ALL
Vehicle Hours Traveled ('000), ALL
Vehicle Hours Delay ('000), ALL
AM PEAK
PM PEAK
MIDDAY
NIGHT
TOTAL
30.9
79,792
2,583
755
26.9
133,369
4,956
1,823
35.8
117,306
3,279
610
43.8
64,827
1,481
86
32.1
395,294
12,298
3,273
AM PEAK
30.4
78,453
2,580
704
Time Period
PM PEAK
MIDDAY
27.4
33.7
127,763
117,554
4,667
3,492
1,526
690
NIGHT
41.7
69,276
1,662
115
TOTAL
31.7
393,046
12,402
3,035
AM PEAK
-2%
-2%
0%
-7%
Time Period
PM PEAK
MIDDAY
2%
-6%
-4%
0%
-6%
6%
-16%
13%
NIGHT
-5%
7%
12%
34%
TOTAL
-1%
-1%
1%
-7%
Sensitivity Results
VMT, VHT and Delay Scenario Comparisons
Base Scenario
10 min CR Headways
1 Extra Lane on all Freeways
15% Higher Freeway Speeds
Doubling Households and Employment
High Auto Operating Cost
VMT
VHT
Delay
393,046
389,881
403,079
399,037
659,835
337,881
12,402
12,222
12,245
12,082
29,110
10,228
3,035
2,929
2,791
3,061
12,181
2,146
Future Work
• Rollout: San Bernardino, North Los Angeles,
Imperial County
• Better methods to handle aggregation bias
• Add highway-only and transit-only
component
• Add more subregion-specific parameters
• Additional optimization
Thank You
Caliper Mapping Software